Merge pull request #15 from TysonAndre/fix-quadratic

Fix quadratic runtime when file has multiple typos

src/LineCounter.php

@@ -0,0 +1,50 @@
+<?php declare(strict_types=1);
+
+namespace PhanTypoCheck;
+
+class LineCounter
+{
+ /** @var string the text used for counting lines */
+ private $line_counting_text;
+ /** @var int the length of the text */
+ private $length;
+ /** @var int the last checked byte offset. */
+ private $current_offset = 0;
+ /** @var int the 0-based line of the last checked offset */
+ private $current_line = 0;
+
+ /** @param array{0:int,1:string,2:int} $token the current token */
+ public function __construct(string $text, array $token)
+ {
+ if (\in_array($token[0], [\T_CONSTANT_ENCAPSED_STRING, \T_ENCAPSED_AND_WHITESPACE])) {
+ // Parse this, but replace `"\n"`, `"\x0a"`, etc. with a single byte character that isn't a literal newline.
+ $this->line_counting_text = StringUtil::parseWithNewlinePlaceholder($token[1]);
+ } else {
+ // There are no escape sequences
+ $this->line_counting_text = $text;
+ }
+ $this->length = strlen($this->line_counting_text);
+ }
+
+ /**
+ * @param int $offset - A 0-based byte offset
+ * @return int - gets the 1-based line number for $offset
+ */
+ public function getLineNumberForOffset(int $offset) : int {
+ if ($offset < 0) {
+ $offset = 0;
+ } elseif ($offset > $this->length) {
+ $offset = $this->length;
+ }
+ $current_offset = $this->current_offset;
+ if ($offset > $current_offset) {
+ $this->current_line += \substr_count($this->line_counting_text, "\n", $current_offset, $offset - $current_offset);
+ $this->current_offset = $offset;
+ } elseif ($offset < $current_offset) {
+ $this->current_line -= \substr_count($this->line_counting_text, "\n", $offset, $current_offset - $offset);
+ $this->current_offset = $offset;
+ }
+ return $this->current_line;
+ }
+
+}

src/TypoCheckUtils.php

@@ -18,11 +18,11 @@
 use function stripos;
 use function strtolower;
 use function substr;
-use function substr_count;
 use function token_get_all;
 use function trim;
 use const PREG_OFFSET_CAPTURE;
 
+require_once __DIR__ . '/LineCounter.php';
 require_once __DIR__ . '/TypoDetails.php';
 require_once __DIR__ . '/StringUtil.php';
 
@@ -86,20 +86,15 @@ public static function getTyposForText(string $contents, bool $plaintext = false
 
  $analyze_text = static function (string $text, array $token) use ($dictionary, &$results) {
  // @phan-suppress-next-line PhanUnusedVariableReference the reference is used to preserve state
- $count_lines_before = static function (int $offset) use(&$line_counting_text, $text, $token) : int {
+ $count_lines_before = static function (int $offset) use($text, $token) : int {
  if ($offset <= 0) {
  return 0;
  }
- if (!isset($line_counting_text)) {
- if (\in_array($token[0], [\T_CONSTANT_ENCAPSED_STRING, \T_ENCAPSED_AND_WHITESPACE])) {
- // Parse this, but replace `"\n"`, `"\x0a"`, etc. with a single byte character that isn't a literal newline.
- $line_counting_text = StringUtil::parseWithNewlinePlaceholder($token[1]);
- } else {
- // There are no escape sequences
- $line_counting_text = $text;
- }
+ static $line_counter;
+ if ($line_counter === null) {
+ $line_counter = new LineCounter($text, $token);
  }
- return substr_count($line_counting_text, "\n", 0, $offset);
+ return $line_counter->getLineNumberForOffset($offset);
  };
  preg_match_all('/[a-z0-9]{3,}(?:\'[a-z]+)?/i', $text, $matches, PREG_OFFSET_CAPTURE);
  foreach ($matches[0] as $match) {